I just started my PhD in an interdisciplinary research lab. For the research I wish to pursue, I am missing some methodological skills, particularly in statistics and machine learning. Are there any suggestions APART from taking online courses? How can I fill this knowledge gap whilst doing all the other (European) PhD duties?
Here is what I did during the beginning of my third year (of 5) in my phd-
deleted all statistical software from my computer save my language of choice (for me it was R). Begin to work on mastering this language.
Watch videos and lectures, get involved in the online statistics community
TAKE CLASSES in advanced stats (though I know this might not be as much of an option in Europe as it was in the US).
go to luncheons with phd students in computational fields
- begin to help people who you can with what you can in stat: nothing helped me learning like taking what I had just watched, practiced, thought about and then applying it to help a lab mate.
- READ, READ, READ (this cannot be understated).
- Practice- learning to code can be frustrating but it is necessary.
- Submit articles using the methods you learned for publication. Getting blind reviewer feedback was extremely helpful for me to see what I was doing wrong.
- attend more classes, continually watch videos, write code daily, publish (or attempt) papers using your methods, practice coding, and READ, READ, READ!
And when I say read, I mean in the medium that is most effective for you. For me, I gained the most from reading journal articles and reading associated blog posts about those journal articles. In this way, if there was something complicated that I did not grasp, the blog would help shore that up in more laymen terms. I did not gain much from reading textbooks, but others do.
Aside from self-educating from available sources (textbooks, peer-reviewed literature, etc.), I would strongly recommend seeking out input/collaboration from research groups that are focused on these areas.
If your doing graduate work on a topic that utilizes these methodologies, chances are you are expected to deliver graduate-level output. This could be challenging when you're outside your element and the resources to do that are not necessarily contained within your own group. This is the place where scientific collaboration is truly fruitful. Input from collaborators will also benefit your work greatly because you can avoid pitfalls that you might encounter alone which may not become apparent until much later.